Potential of multi-source Geospatial data in Accurately Estimating the Live Storage Capacity of Reservoir

Abstract. The two-fold pressure of ongoing rise in water demand and the impacts of prevailing warming climate are likely to exacerbate the shrinking of major water-bodies, especially in arid/semi-arid regions. Presently, assessment of live storage capacity (LSC) of any gauged reservoir needs water level observation and updated Area-Elevation-Capacity curve, whose concurrent availability is challenging for water managers. Recent studies highlight satellite altimetry as a game-changer for estimating water elevation and lake storage, especially useful for remote, inaccessible regions, where traditional gauging stations are scarce. Yet, these measurements carry inherent limitations due to data capture procedures. The motivation of the present study originates from exploring the benefits of using geospatial data, furthering efforts to reduce a number of limitations that causes uncertainties in estimating lake storage. This study aims to develop a novel geospatial-based methodology to remotely estimate the live storage capacity of reservoir and further limited by data availability. This methodology is based on trapezoidal rule using Area-Height relation, where incremental live storage capacity estimates of infinitesimally small layers between consecutive water levels (likely to vary between minimum drawdown level and maximum full reservoir level), culminates into a running cumulative reservoir storage capacity. The satellite imagery-based water-surface area and corresponding water elevation dataset were assessed for at least one water-year. Initial trials with field data were later replaced by altimetry data, and the latter was validated to make the methodology purely geospatial data based. The geospatial based LSC estimates of Ukai reservoir (Gujarat, India) gave an excellent match to the level of 5th decimal digit of accuracy (mean error 8.79e-06, standard deviation of error 9.11e-05, RMSE 9.10e-05 MCM, Bias 8.79e-06) when compared to that of observation-based estimates. The fusion of optical and SAR imageries reduced the revisit gaps and missing data points, especially during monsoon, thus allowing a temporally richer dataset of water surface area for the study period. In contrast, only single altimetry data source (27 days revisit) was available for Ukai reservoir. This coarse water elevation data was pre-processed using advanced cubic spline interpolation method to obtain regular, temporally-richer time series, and was validated. It is concluded that a novel methodology is developed that (a) can accurately estimate LSC of any gauged reservoir; (b) the estimated accuracy of LSC is mainly a function of availability of temporally-rich geospatial input (water surface area and validated water-level), and an apt computation platform allowing complex integral computation among other factors for accurate estimation.